By: Johan Potgieter - Cluster Industrial Software Lead at Schneider Electric

What is machine vision? By definition, it allows automated systems to “see” the real world; an ocular implant for machines, if you will.  And like Ironman’s heads-up display, machine vision uses cameras, sensors and software to capture and analyse images from various applications, optimising decision making.

Machine vision can be traced back to the 1950s and 1960s, when scientists began exploring ways for computers to interpret visual data. One of the earliest projects was led by Marvin Minsky at MIT in 1966, which aimed to teach a computer to recognise simple objects

In recent years machine vision has undergone significant advancements, most significantly 4D perception.  Unlike traditional 3D vision, which captures spatial data, 4D perception adds the dimension of time, allowing systems to understand the movement and speed of objects.

This remarkable innovation plays a vital role in dynamic settings where real-time decision making is paramount. For example, 4D Light Detection and Ranging (LiDAR) technology integrates complex sensors onto a single chip, providing high-resolution spatial and temporal data. In turn, it enhances the ability of robots to navigate and interact with their surroundings.

4D machine vision at work

Machine vision’s 4D capability combines spatial and temporal data to create a comprehensive understanding of an environment. Not only do these advanced sensors and algorithms capture the position and shape of objects, but also movement over time. 

By analysing this data, systems can then predict future positions and interactions, enabling more precise and adaptive responses. This is particularly useful in applications like autonomous driving, where understanding the speed and trajectory of surrounding vehicles is critical for safety.

In robotics and autonomous systems

On the factory floor, 4D perception enables robots to perform complex tasks such as such as object recognition, pose estimation, and depth perception with exceptional accuracy.

Moreover, it enables robots to adapt to changing environments, improving their operational efficiency and reducing the likelihood of errors. Additionally, 4D perception facilitates better coordination between multiple robots, enhancing collaborative tasks and streamlining production processes.

Today, machine vision is used in numerous industries and include:

•                    Quality control and inspection - machine vision systems are widely used in manufacturing for quality control, ensuring products meet specified standards by detecting defects and inconsistencies.

•                    Autonomous vehicles - self-driving cars use machine vision to navigate roads, detect obstacles, and make real-time driving decisions.

•                    Healthcare - in medical imaging, machine vision aids in diagnosing diseases by analysing medical scans and detecting anomalies.

•                    Agriculture- machine vision helps in monitoring crop health, detecting pests, and optimising harvests.

•                    Retail - automated checkout systems use machine vision to scan and recognize products, streamlining the shopping experience.

At Schneider Electric, we offer high-quality and optical filters to optimise machine vision, ensuring precise detection and analysis of objects. Furthermore, our AVEVA Vision AI Assistant is a powerful tool that leverages deep learning to enhance machine vision capabilities. 

Integrated with the AVEVA HMI (human-machine interface) and SCADA (supervisory control and data acquisition) software, this assistant monitors real-time image streams and alerts operators to anomalies, increasing situational awareness and preventing machine failures. 

The Vision AI Assistant can be used for various applications, including quality control, maintenance, and operational efficiency.